University of Wisconsin-Madison Institute for Research on

Volume 13

Number 2

Summer 1991

The : Assessing what we have learned 1 Small grants IRP agenda for 1991-1993: Education, families, and Mentoring of at-risk students 26 IRP summer research workshop: Problems of the low-income population 33 Poverty research seminars 34 ISSN: 0 195-5705

The underclass: Assessing what we have learned by William R. Prosser

class behavior and concentration. Because it was felt that William R. Prosser is a Senior Policy Analyst in the Office these papers would not necessarily provide a unifying con- of Human Services Policy, Office of the Assistant Secre- ceptual base to guide policy discussions to deal with the tary for Planning and Evaluation, U.S. Department of underclass, four theoretical models were commissioned the Health and Human Services. The opinions expressed in this following year. article do not necessarily represent the views of the Depart- ment or the Bush administration. Prosser will be a visiting In March 1990, ASPE and the Joint Center for Political and in the School of Social Work, University of Wis- Economic Studies (JCPES) cosponsored a forum at which consin-Madison, during the 1991-92 academic year. these papers were presented and discussed. The conference was organized to follow a series of smaller conferences and workshops covering various aspects of the underclass, sponsored over a period of several years by JCPES. It also followed a much more comprehensive conference on the Introduction "truly disadvantaged," jointly sponsored by the Social Sci- ence Research Council and the Center for Urban Affairs Over the last five years, policy and scholarly debates have and Policy Research at Northwestern University in October focused a great deal of attention on a group called the 1989.' "underclass." During 1987 and 1988, the Office of the Assistant Secretary for Planning and Evaluation (ASPE) of The series of conferences and attendant papers were a the U.S. Department of Health and Human Services com- significant investment in an important line of research on missioned ten papers on the underclass in hope of inform- low-income people and the neighborhoods in which they ing this debate. The six papers commissioned in 1987 were live. This article is an attempt to synthesize what we have empirical explorations of various issues related to under- learned from ASPE's share of the research. While it deals primarily with the conference papers and discussions, it Wilson is now promoting a definition of the underclass also draws on a few scholarly materials which have become developed by Martha Van Haitsma, who defines the under- public since then. This synthesis is motivated by a belief class as "those persons who are weakly connected to the that one of ASPE's primary roles is to try to inform the formal labor force and whose social context tends to main- public debate on important social issues and to enhance tain or further weaken this atta~hment,"~a definition that public policy resear~h.~ requires an explanation of the terms "labor force attach- ment" and "social context." In her conceptual definition of ASPE's general interest in research on the underclass was "labor force attachment," Van Haitsma uses a structural reflected in the following questions, raised in the solicita- concept, stability, or the likelihood that the income will tions of research studies and conference agenda: continue, and a normative concept, the social acceptability of the work. She defines "social context" as the household, Is "underclass" a useful concept? neighborhood, and social network in which the individual What have we learned about the underclass? is embedded. Does the concept of neighborhood concentration bring Her definition places a heavy emphasis on low income as a anything new to the discussion of poverty? How do result of low labor force participation and the structural neighborhoods and neighbors influence individual forces that lead to low participation. In effect, she subordi- behavior? nates such behaviors as dropping out of school, teenage What have we learned that is policy relevant? What childbearing, criminal activity, and to does the research tell us that could lead to policy their impact on labor force participation and the income prescriptions? derived from it. Work outside the home is the only outcome valued in her theory. (Work in the home-parenting-is In this article, I examine what we have learned in answer to missing from her equation.) these questions. I look in turn at definitions, paradigms used for analyses, findings, avenues for future policy re- Other scholars, notably Erol Ricketts and Isabel Sawhill, search, and the question of whether policy prescriptions have approached the problem of conceptualizing (or flow from what we have learned. operationalizing Wilson's definition of) the underclass with even less emphasis on poverty. Instead they use a cluster of behaviors (and assumed attitudes) that are con- sidered outside today's middle-class social norms, accord- Definition of the underclass ing to which young people complete high school, delay childbearing until they are able to support their offspring, work or are supported by a spouse, and obey the law.7 The current discussion of the poor and the under- class started in the early 1980s. The fact that recent debate Poverty is highly associated with the underclass because it was initiated by a journalist, Ken Auletta, may account for is related to the structural and behavioral variables used. some of the ambiguity about its definition.' The "under- Poverty may be temporary or persistent over a person's class" is certainly related to Karl Marx's "lumpen prole- lifetime or across generations. It is generally accepted, tariat," Edward Banfield's "lower class," Michael Hamng- however, that not all the poor are members of the under- ton's "other America," Oscar Lewis's "," class and not all the members of the underclass are poor. and Elliot Liebow's "Talley's ~orner."~ Furthermore, the behavioral norms used vary over time and from one social group to another. Out-of-wedlock defines the underclass as childbearing is a good example: the incidence has changed that heterogeneous grouping of families and individuals dramatically over the last several decades. Births to unmar- who are outside the mainstream of the American occu- ried women are now commonplace among some groups pational system. Included . . . are individuals who lack while remaining outside the social norm for others. training and skills and either experience long-term un- employment or are not members of the labor force, Another way of looking at a particular behavior associated individuals who are engaged in street crime and other with the underclass at a static point in time, such as out-of- forms of aberrant behavior, and families that experience wedlock childbearing, is the probability that it is related to long-term spells of poverty and/or welfare dependen~y.~ other factors, such as lack of education, , or low income; and these in turn are related to both individual Wilson has done much of the seminal thinking on this topic. choices and larger community structural force^.^ Unfortu- He is at present carrying out a very large study, Urban nately, our measurement techniques at present are not up to Family Life (UFL) in Chicago, supported in part by ASPE. the task of determining primary causes of particular behav- The project is probably the most comprehensive effort yet iors, so we cannot yet separate those who are part of the undertaken to study the underclass. It includes a household structural underclass through no fault of their own (the poor survey, ethnographic work, and studies of institutions and deserving of concern and support) from those whose own administrative data. actions bring about their categorization (the undeserving poor). It could be the case that for most members of the underclass, their choices are so limited by structural factors beyond their control-poor parents, poor schools, lack of FOCUS is a Newsletter put out three times a year by the jobs, no marriageable men, etc.-that for all intents and purposes they have no choices to make. This too begs the Institute for Research on Poverty point, since not all people born into the same terrible ghetto 1180 Observatory Drive situations choose underclass behaviors. Trying to place 3412 Social Science Building individuals into "deserving" and "undeserving" pigeon- University of Wisconsin holes, however, is a futile exercise because life is not static; Madison, Wisconsin 53706 it is a string of situations and choices with varying degrees (608) 262-6358 of freedom and determinism. The Institute is a nonprofit, nonpartisan, university-based The definitions suggest that two broad social science camps research center. As such it takes no stand on public policy are involved in this discussion: structuralists and issues. Any opinions expressed in its publications are those behavioralists. The structuralists, such as Wilson and of the authors and not of the Institute. Douglas Massey, who tend to be sociologists, have defined the issue in terms of broad societal forces that are causing The purpose of Focus is to provide coverage of poverty- neighborhoods to deteriorate and leaving their inhabitants related research, events, and issues, and to acquaint a large minimal opportunities to achieve a reasonable standard of audience with the work of the Institute by means of short li~ing.~Sometimes they have left it somewhat ambiguous essays on selected pieces of research. A subscription form how their theories could be tested empirically.1° with rates for our Discussion Papers and Reprints is on the back inside cover. Nonsubscribers may purchase individual Behaviorialists, such as Sawhill, Ricketts, Ronald Mincy, papers from the Institute at $3.50 for a Discussion Paper and others, who tend to be economists (although Ricketts is and $2.00 for a Reprint. a sociologist), have focused on individuals and categorized them on the basis of groups of behaviors. This work has Focus is free of charge, although contributions to the U.W. been descriptive in nature and empirically based for the Foundation-IRP Fund sent to the above address in support most part; but the behavioralists have not articulated an of Focus are encouraged. overarching theory beyond the standard economic theory of utility maximization under uncertainty and constraints. Edited by E. Uhr.

Discussion of the underclass is further confused by differ- Copyright O 1991 by the Regents of the University of ences between definitions used for theoretical, journalistic, Wisconsin System on behalf of the Institute for Research and rhetorical purposes, and operational definitions used to on Poverty. All rights reserved. study the underclass or empirically test hypotheses about it. Generally, the operational definitions have been arbitrary, driven by the available data. For example, although crimi- nal activity and drug use are often associated with the class-weak labor force attachment, dependency on wel- underclass, Ricketts and Sawhill did not use them in their fare, teenage pregnancy, dropping out of school, and crimi- analysis of underclass areas, because there are no measures nal activity; and (3) the persistence of these behaviors of these activities in the decennial data sets they used. In across two or more generations.14 That is, the underclass is addition, their unit of analysis was not an individual but a associated with large numbers (or percentages) of people geographic area-a census tract-where a significantly with one or more dysfunctional characteristics living in higher proportion than average of all the factors were particular areas, sharing their dysfunctional traits with their present. parents or having them for an extended period, or having several of the behaviors simultaneously during their life- Scholars also have differed in their means of measuring the time. "Underclass," however, is no more precise than extent to which a neighborhood may be categorized as "," as in "middle-class values," or "rich," as in underclass. Some measure the proportion of individuals in "soak the rich," an expression heard during the 1990 tax poverty ," others use a significant deviation from a number debates. The underclass appears to be, like beauty, in the of behavior norms in an areal designation such as census eye of the beholder: People do not know how to define it tract or a zip code.I2 Still others prefer the boundaries precisely, but think they know it when they see it. defined by the inhabitants of the neighborhood.13 None of the papers examined the trends in the growth of the In general, we see that the underclass has been defined in underclass, but other works have shown that, however de- three ways: (1) a geographic concentration of individuals fined, the underclass increased from 1970 to 1980, and in with some characteristic associated with the underclass, 1980 between one and two million people could be charac- such as poverty; (2) common occurrence in a given locale terized as members of the underclass living in underclass of several forms of behavior associated with the under- areas.15 How an underclass person or neighborhood is defined, were lower yet for whites (1 percent or less), using areal or however, dramatically affects estimated prevalence. For personal outcomes. These data show that no matter how example, Table 1, prepared by Hill and O'Neill (paper 3), you define the underclass (by area or behavior), on average, presents data from the National Longitudinal Survey of many underclass people are living next door to neighbors Youth (NLSY) showing the percentage of youth who meet who cannot be so categorized, and the more traits used several underclass neighborhood or behavioral defini- simultaneously to define members of the underclass, the tions.16 The proportion of young men and women of differ- rarer they become. These data also show the very different ent racial-ethnic categories living in underclass areas var- experiences of whites, blacks, and Hispanics. Whites have ied significantly by the number of variables used and by a lower prevalence of underclass behaviors and are much whether the person was white or not. Approximately 40 to less likely to live in underclass areas. 45 percent of blacks and Hispanics lived in areas of high welfare concentration. For whites it was about 5 percent. The percentages of black, Hispanic, and white youth living in areas of high poverty concentration were about one-half Conference discussion the figures for welfare concentration. Lack of agreement of definitions of the underclass, failure When an operational definition used three or four variables to operationalize theoretical statements, and differing per- simultaneously to classify an underclass person on the basis spectives have led to significant communication problems of personal behaviors (time out of the labor force, failure to within the scholarly community and between the research complete high school, and time in jail), or an area, based on and public policy communities. The conference gave ample high concentration of people with underclass characteris- evidence of these problems from the opening remarks tics, the figures dropped dramatically, to about 3 percent through and including the wrap-up. Martin Gerry, the As- for blacks. For Hispanics the proportions were slightly sistant Secretary for Planning and Evaluation, during his higher in areal concentrations (about 4 percent), and lower introductory remarks addressed two concerns: blaming the for personal outcomes (about 1 percent). The percentages victim and statistical .

Table 1 Prevalence of Underclass Status among Youth in the NLSY Living with Their Families and in SMSAs in 1979

Men Women Underclass Definition Black Hispanic White Black Hispanic White

Zip rode characteristics % in areas with a high concentration ofa Welfare families Families in poverty Ricketts-Sawhill characteristics Personal outcomes % men with A high proportion of low-work yearsb And a high school dropout And a high school dropout and ever in jail % women with A high proportion of years with low work and welfare receiptC And high school dropout And high school dropout and teen mother

Source: M. Anne Hill and June O'Neill, "Underclass Behaviors in the United States: Measurement and Analysis of Determinants," City University of New York, March 1990. (Paper 3.) a These areas are neighborhoods scoring a standard deviation above the weighted mean in the characteristic indicated. The Ricketts-Sawhill measure counts an area as underclass if it scores high on the incidence of welfare families, female household heads, high school dropouts, and men out of the labor force (see Erol Ricketts and lsabel Sawhill, "Defining and Measuring the Underclass," Journal of Policy Analysis and Management, 7 [Winter 19881, 3 16-325). "Low-work years" refers to youth who worked fewer than 26 weeks per year for more than 42.4 percent of potential years since age 19, when they were not enrolled in school. not in the army, and not in jail. "Low work and welfare receipt" refers to women who received welfare for more than two months and worked fewer than 26 weeks for more than 19.8 percent of potential years since age 19. I come to today's discussion with an open mind, tem- On the positive side, however, Jencks also pointed out that pered of course by my own experience, but concerned study of the underclass is bringing together scholars who about the growing tendency to use the concept of the have tended to compartmentalize themselves into their own underclass to "pathologize" individuals and families in little cottage industries, carrying out research about single poverty. While there may be important insights gained dependent variables associated with only one of the pat- from projecting characteristics onto a group based on terns of behavior used to discuss the underclass+.g., wel- aggregate data, it is equally important to recognize the fare dependency, weak labor force attachment, crime. Re- wide variety of individual patterns related to such char- search on the underclass has encouraged people to look at acteristics. interactions of these behaviors and whether their effects are even more pronounced when geographic concentration is "Blaming the victim" suggests that people use "underclass" incorporated into the analysis. to label groups of people pejoratively on the basis of dys- functional behavior over which the people so labeled have no control. "Statistical discrimination" occurs when one What have we learned from this line of assumes that the underclass is very homogeneous and quite research on the underclass? distinct from the nonunderclass. Individuals are then classi- fied as belonging to the underclass on the basis of a small Short abstracts of the ten conference papers are presented in number of characteristics, on their address, or their appear- the accompanying box. This section attempts to synthesize ance.'' Kathryn Neckerman and Joleen Kirschenman report hypotheses which seem to emerge from these ten papers an example of this problem when employers stereotype and the conference discussion. It is organized around the particular applicants for jobs on the basis of assumed group primary themes used to define the underclass: neighbor- behavior.l8 The fuzzy nature of the underclass concept may hood concentration of people with underclass behaviors, contribute to both blaming the victim and statistical dis- persistence of an underclass behavior, and specific patterns crimination. of behavior.

Herbert Gans expressed these same concerns while partici- Neighborhood concentration of underclass members pating in the conference, as well as in several papers.I9 Does "neighborhood" and/or "concentration" add to our understanding? How might concentration of those with Wilson recently discussed concerns about "blaming the underclass characteristics (however defined) come about? victim." He devoted a substantial portion of his presidential Is it related to in-migration of underclass individuals or out- address at the 1990 annual meeting of the American Socio- migration of the nonunderclass? Does level of concentra- logical Association to responding to Gans. He cautioned tion affect behavior, values, attitudes, or preferences of the against a situation similar to the period after the Moynihan residents in the neighborhood? If so, how? Are individual Report on the black family was published in the middle racial-ethnic groups affected differently? Does a neighbor- 1960s, when some scholars shied away from poverty re- hood just gradually get worse as the percentage of residents search to protect themselves from charges of racism.20 with underclass behaviors increases; or does it suddenly get significantly worse when the percentage or number of Summing up the conference, Christopher Jencks addressed underclass residents reaches a critical mass? the lack of definitional clarity. He expressed the opinion that the concept of underclass, like class, is useful in some Three of the theoretical papers deal with these issues. All situations and not others. It is useful in lunchroom discus- three are written from the behavioralist perspective. sions. It gives one a sense of ordering individuals or fami- lies along some composite measure that has to do with Jorge Martinez-Vazquez and Rubin Saposnik (paper 9) hy- financial, human, social, and cultural capital. When it pothesize that when income opportunities can come from comes to empirically testing hypotheses, we must be pre- both conventional (legal) and unconventional sources, cise in our definitions. Disagreements about operationaliz- when utility is a function of real income and psychic in- ing variables for empirical work should not be seen as come related to one's social standing with others, and when damaging the value of the theoretical formulations. the price to the individual of conventional or unconven- Fuzziness is not necessarily harmful; it depends on the tional behavior is inversely related to the local concentra- situation and the alternatives. tion of that sort of behavior-then, underclass behavior, however defined, is more attractive the higher the concen- Jencks also reiterated a caveat that has motivated his writ- tration of underclass individuals in the neighborhood. ings in this area: we should not talk about the various underclass behaviors as if they all have the same causes and Regarding criminal behavior, if police resources are rela- consequences. "Nobody really believes that the process of tively fixed and the probability of being apprehended getting pregnant is the same as the process of mugging decreases with the number of criminals operating in a s~mebody."~'Furthermore, these phenomena call for very different policy prescriptions. (Text continued on p. 9) Conference Papers

1990 ASPE-JCPES Conference on the Underclass

To obtain copies of these papers, contact the authors or the Policy Information Center, Office of the Assistant Secretary for Planning and Evaluation, U.S. Department of Health and Human Services, (202) 245-6445.

Empirical studies

1. Mary Corcoran, Roger Gordon, Deborah Laren, welfare reflects tastes for welfare passed on from gen- and Gary Solon, "Effects of Family and Community eration to generation. Three questions are addressed: (1) Background on Men's Economic Status," University How large is the correlation between mother's and of Michigan, February 1989. (This is the paper that daughter's welfare experience? (2) Does the relationship was delivered at the conference.) reflect similar constraints or similar tastes? (3) Does evidence exist to show that this correlation reflects a This study uses intergenerational data from the Panel causal relationship; does the act of accepting welfare Study of Income Dynamics (PSID) to investigate the alter the tastes that are transmitted to the next genera- effects of family and community background on men's tion? The PSLD is used to generate descriptive tables and economic status. It is distinguished from most previous to estimate the model. The analysis confirms an studies by its emphasis on a comprehensive set of com- intergenerational relationship between welfare use of munity influences and on influences from poverty and mothers and daughters. It leaves unanswered the reason welfare use. It finds substantial disadvantages in eco- for the association. Is the daughter's behavior an effect nomic status for black men, men from lower-income of the mother's attitudes and actions, or is there a corre- families, and men from welfare-dependent families or lation with other variables that both share which have communities. Otherwise, not much evidence of commu- not been studied, or are there other explanations? nity influences is found. The lack of results may be due to the grossness of the geographic measures of commu- nity variables. 3. M. Anne Hill and June O'Neill, "Underclass Be- haviors in the United States: Measurement and Mary Corcoran, "Problems of the Underclass: Analysis of Determinants," City University of New Underclass Neighborhoods and Intergenerational York, March 1990. Poverty and Dependency," University of Michigan.- (This paper was commissioned but will not be available This project develops measures of the incidence of until the fall of 1991.) underclass behaviors among a sample population and applies various multidimensional indices of underclass This project examines how community disadvantages status in classifying individuals. It measures how this hinder children's future economic prospects. It uses data incidence varies by race, ethnicity, geographic area, and from the PSID augmented with data on neighborhood whether the individual came from an underclass back- characteristics measured at the census-tract level. It in- ground. The study uses data from the Youth Cohort of vestigates eight family and neighborhood factors (edu- the National Longitudinal Survey of Youth (NLSY) and cation, labor supply, timing of marriage, timing of fertil- matches it with census zip code file data. It attempts to ity, marital instability, welfare dependence, number of identify the important specific determinants of under- children, and socioeconomic structure of the community class status, whether defined by residence in a neighbor- in which the young adult lived after leaving home) that hood with a high concentration of individuals with cer- affect young adults' economic outcomes. A sorting tain characteristics or by classifying individuals with a model, an incremental model, and an epidemic model, cluster of certain behaviors. It examines several broad all of which illustrate the causes and consequences of categories of explanatory variables in relation to many increases in concentrations of poverty and other social kinds of underclass behaviors: weak labor force attach- ills, are examined. ment, welfare dependence, out-of-wedlock childbear- ing, criminal involvement, and low educational attain- 2. Peter Gottschalk, "Is Intergenerational Correla- ment. Among various speculations about the important tion in Welfare Participation across Generations determinants of underclass behavior, two are tested: the Spurious?" Boston College, November 1990. Wilson hypothesis~. and the Murray. hypothesis.~. William Gottschalk focuses on the perceived relationship in wel- Julius Wilson proposes that lack of labor market oppor- fare status between mothers and daughters. A presump- tunities is responsible for low rates of labor force attach- tion exists that the propensity for a daughter to receive ment among minority men. Charles Murray proposes Conference Papers, Cont. that the availability of welfare benefits is the dominant from 1985 to 1987. Whites are found to be most likely to factor responsible for low rates of labor force attachment choose abortions, followed closely by blacks. Latinos among minority men. are far less likely to choose abortion. Nearly three- fourths of Medicaid recipients choose not to have abor- 4. Shelly Lundberg and Robert Plotnick, "Testing tions, while more than three-fourths of those not receiv- the Opportunity Cost Hypothesis of Teenage Out-of- ing Medicaid choose abortion. In response to teenage Wedlock Childbearing," University of Washington, pregnancy, most of the fathers in the predominantly 1990. white and Latino communities marry after the preg- nancy is discovered but before the child's birth. The This study addresses the causes and consequences of black community chooses not to marry. Latinos and premarital childbearing among adolescents. The report blacks are less likely to marry than whites, and blacks estimates the effect of early childbearing on adult are less likely to marry than Latinos. AFDC levels are wages, both within and outside of marriage, relative to about 50 percent for the two minority neighborhoods remaining childless as a teenager. It also estimates the and 10 percent for the white neighborhood. The study effects of teenage fertility and marital outcomes on earn- concludes that (1) the white neighborhood's low rates of ings. Several measures of welfare benefits are devel- AFDC enrollment are due to the young fathers' abilities oped, which affect the net income losses associated with to find jobs and apartments for their families; (2) the premarital childbearing. In addition, various indices of lack of marriage in the black community allows fathers the costs and availability of abortion and family plan- to supplement the AFDC payments to their families, but ning services are developed. The primary data base used these arrangements tend to be highly unstable; (3) the for the study is the NLSY. Welfare policy, abortion, and need for AFDC in the Latino community relates to high family planning variables are gathered from published levels of poverty and lack of access to employment but sources and appended to the NLSY records. Because conflicts with cultural emphasis on marriage and legal race and ethnic differences in rates of premarital paternity, which makes AFDC eligibility more difficult. childbearing are large and of interest, separate analyses are conducted for whites, blacks, and Hispanics. Alter- native fertility and marriage choices during teen years 6. Mark Testa and Marilyn Krogh, "Nonmarital Par- are found to affect future wages and mamage probabili- enthood, Male Joblessness and AFDC Participation ties: (1) For a white female adolescent, having a child as in Inner-City Chicago," University of Chicago, 1990. a teenager reduces long-term potential earnings by 10 This report uses data from the 1987 Urban Family Life percent to 15 percent. (2) For a black female adolescent, project directed by William Julius Wilson to study having a child as a teenager, whether premaritally or nonmarital parenthood by examining initial marital, pa- postmaritally, tends to increase potential earnings by 10 rental, and welfare transitions of people ages 18 to 44 in percent to 30 percent. The estimates for Hispanics are inner-city Chicago. Findings include the following: The unstable, possibly as a consequence of the small sample risks of becoming a parent are the same for stably em- sizes in the early childbearing groups, and could not be verified by this analysis. The study finds mixed results ployed and chronically unemployed black males, but as to whether potential opportunity costs influence teen black women's work status and school status strongly fertility choices. affect their risks of becoming premaritally pregnant. Black men who are stably employed are twice as likely 5. Mercer L. Sullivan, "Patterns of AFDC Use in a to marry as black men who are not in school, not in the Comparative Ethnographic Study of Young Fathers military, and not at work. The decline in marriage and Their Children in Three Low-Income Neighbor- among inner-city blacks is not due simply to increases in hoods," Vera Institute, June 1990. the proportion of black males who are jobless; rather, it is concentrated among the chronically jobless. Racial This study examines the relationships between minority differences in male employment status, however, cannot status, , abortion and marriage de- account for the racial differences in postconception mar- cisions, out-of-wedlock childbearing, and enrollment in riage rates in inner-city Chicago. Finally, mothers who AFDC. Young fathers were interviewed in three pre- are married at the birth of their first child are signifi- dominantly white, Latino, and black low-income neigh- cantly less likely to receive AFDC than mothers who are borhoods in Brooklyn, New York. Data from the 1980 unmarried, but these effects even out when marital dis- census were examined; and abortion and birth records solutions that occur after the child's birth are taken into were analyzed of teenage females living in these areas account. Conference Papers, Cont.

Theoretical studies

7.William A. Darity, Jr., Samuel L. Myers, Jr., Wil- 9. Jorge Martinez-Vazquez and Rubin Saposnik, "A liam J. Sobol, and Emmett Carson, "How Useful Is Contagion Model of Underclass Neighborhoods," the Black Underclass? Critical Essays on Social Sci- Georgia State University, 1990. ence and Methodology," University of Maryland, This research investigates theories regarding the infec- November 1990. tious aspects of underclass neighborhoods. The paper Literature reviews and economic analyses are used to presents a basic behavioral model and both analytical examine (1) structural versus behavioral explanations and simulation solutions for the growth of the underclass for the growth of the underclass; (2) deviant behavior in a single neighborhood. The model illustrates how and crime; (3) historical considerations of family insta- areas can become underclass neighborhoods and how bility among blacks; (4) the effects of welfare and crimi- underclass behavior may affect all individuals in the nal on female-headed families; (5) the role of neighborhood. The paper also attempts to show how the courts in shaping the family structure of black families growth of the underclass depends on underlying socio- in America; and (6) the usefulness of the underclass economic parameters; and it analyzes the impact of mi- concept for public policymaking. The study finds that gration on the concentration of the underclass in particu- behaviorally based notions of the black underclass are lar neighborhoods. (This study also includes an not useful in understanding family instability and crime; extensive review of the underclass literature, authored racial classifications may be more appropriate for under- by David L. Sjoquist.) standing specific aspects of underclass behavior. Central to an understanding of the rise in female-headed families 10. Douglas A. Wolf, Rebecca Clark, and Vicki and the escalation of violent crime among blacks are the Freedman, "Modeling the Growth of the Underclass: diminishing value of black labor and the marginalization Neighborhood Effects and Neighborhood Dynam- of blacks located at the bottom of society. The un- ics," Urban Institute, May 1990. wantedness of blacks at the bottom of the ladder mani- This study examines how a person's neighborhood in- fests itself in declining labor force participation, increas- fluences hisher decisions. Models of underclass behav- ing criminal violence, withdrawal from the productive ior are formulated. The behavior includes crime, welfare spheres of life, reductions in the supply of marriageable dependency, joblessness, teenage pregnancy, child males, and increases in female headship. abuse, and other dysfunctional activities. A review of the literature on the topic reveals that research has made 8. Scott Freeman, Jeffrey Grogger, and Jon little progress in specifying how neighborhood effects Sonstelie, "The Concentration of Crime and Under- operate. The decision to drop out of high school is then class Neighborhoods," University of California, used to propose three mechanisms through which a Santa Barbara, January 1990. person's neighborhood shapes hisher decisions: This study addresses reasons why crime concentrates in 1. The informational model. Observation of the actions certain neighborhoods even though adjacent neighbor- taken by others, and of the consequences of those ac- hoods may be crime free; and why crime concentrates in tions, leads to the formation of expectations regarding poor neighborhoods rather than in wealthy areas where the consequences of one's own actions. potential rewards would appear greater. Attention is 2. The preference formation model. The more prevalent given to the way the choices of individuals to work or nonnormative behavior is in one's immediate social set- engage in crime give rise to underclass areas. The report ting, the higher a person ranks such behavior in his own examines a model of the distribution of crime between preference structure. two hypothetical communities with originally identical 3. Rewards and sanctions model. One's taste for alter- populations, given fixed police expenditures. The model nate behaviors is not influenced by the actions of others, offers an explanation for the rapid deterioration of some but one's costs and benefits of taking those actions are. neighborhoods. It demonstrates how the recessions of the 1970s (which lowered wages) and crack cocaine The study finds the strongest support for the first model. trade in the 1980s (which increased the returns to crime) It also points out that if neighborhoods affect people's may have greatly accelerated the development of under- decisions, then people's decisions help define the con- class areas. Differing assumptions are applied to the text in which the decisions of others are made. model to encompass many circumstances. (Continuedfrom p. 5) they use multivariate analysis to look at differences in outcomes for youths living in and outside areas with high neighborhood, then Scott Freeman, Jeffrey Grogger, and concentrations of welfare families (at least one standard Jon Sonstelie (paper 8) hypothesize that (1) crime will deviation above the national mean for zip codes), they find concentrate in poor neighborhoods; (2) victims and perpe- that living in areas with welfare concentration is signifi- trators will both generally be poor; and (3) noncriminal cantly related to little work experience, poor achievement residents of those neighborhoods will be less likely to work scores, and highest grade completed for white men, but not and invest in education than their counterparts in other black men. Welfare concentration is found not to be related neighborhoods. Furthermore, they predict that if neighbor- to other dependent variables for the men, such as school hoods that have been segregated by race-ethnicity (which is dropout, having been in jail, drug use, or having fathered a not correlated with ability) are opened up through a reduc- child out of wedlock. For the young black women (but not tion in discriminatory housing practices, then higher-abil- young white women), living in a high welfare area is sig- itythigher-income residents will tend to move out, even nificantly related to out-of-wedlock first births, high school after paying rent premiums for crime-free areas. In addi- dropout, having received Aid to Families with Dependent tion, their model predicts that when housing stock is vari- Children (AFDC), and a large proportion of years during able, the differential between rents in crime-ridden and which they worked little and were heavily dependent on crime-free areas will lead to a decline in housing stock in AFDC. the crime-ridden areas. Harry Holzer, one of the discus- sants, questioned the hypotheses and concluded that the Mary Corcoran, Roger Gordon, Deborah Laren, and Gary relative magnitude of these effects would be minor com- Solon (paper 1) also look at the influence of family use of pared to other factors.22 welfare and neighborhood concentration of welfare fami- lies on young men's economic status-hours worked and Douglas A. Wolf, Rebecca Clark, and Vicki Freedman wage rates. They report that their impression is that after (paper 10) discuss the mechanisms that might operate if controlling for other family and community characteristics, concentrated behavior is more influential on individuals the presence of either family welfare receipt or high com- than dispersed behavior. They present three possibilities: munity welfare participation rates is associated with nega- an information model whereby observation of others' ac- tive economic status variables for young men-lower wage tions and resulting consequences influences one's own ex- rates and incomes. Other community variables (median pectations; preference formation, which is an imitative, income, male unemployment rate, and percentage of fami- adaptive evolution of tastes based on the prevalence of peer lies that are female headed with children) generally show behavior in one's immediate social setting; and a rewards negligible associations with the wage rates, hours worked, and sanctions model, where the personal costs and benefits and income of the young men. In this work, Corcoran and of one's actions are influenced by the behavior of others- her colleagues use fairly gross spatial measures, which are e.g., peer pressure. They mention that in measuring these now being refined. Further, they acknowledge that un- influences, researchers must be cognizant of whether their avoidable measurement error and omitted variables make it variables measure influences close to the individual's impossible to determine from this research whether there home-neighborhood or peers-or more global influ- are welfare "effects." ences-metropolitan or larger geographic forces. They also point out that if neighborhoods affect people's behavior, None of the empirical papers addresses the issues of racial then there is a reciprocity such that one's behavior exerts an concentration in ghetto areas or the disadvantage that mi- influence on others. They conclude that these are impor- norities may suffer from spatial mismatch, as jobs move to tant, testable formulations for future rational choice models the suburbs, or the skill mismatch caused by the shift from though they have not performed empirical verification2' manufacturing to service industries. William A. Darity, Jr., Samuel L. Myers, Jr., William J. Sobol, and Emmett Their paper also links their theoretical formulation to the Carson (paper 7) along with Wilson; Massey, Eggers, and work of Charles F. Manski and Peter A. Stre~fert.~~If the Denton; Kasarda; and others believe that racial factors con- underclass consists of those who remain in a neighborhood tribute greatly to the development of the ~nderclass.~~ when others migrate out, this selection process may lead those who remain to have a negative in their perception of opportunities for success. The underclass, as a result of selective out-migration, may underestimate the opportuni- New Institute Director ties for and rewards to more middle-class behavior.

The empirical work of M. Anne Hill and June O'Neill Robert M. Hauser, Vilas Professor of Sociol- (paper 3) bridges the work of structuralists and behavior- ogy at the University of Wisconsin-Madison, alists. In addition, it uses both spatial concentration and became Director of the Institute for Research clusters of behaviors to analyze the underclass. A number on Poverty, July 1, 199 1. of explanatory variables are explored in their analyses- individual, family, neighborhood, city, and state. When It appears that at least one type of neighborhood concentra- women whose mothers had more education have higher tion, people on welfare, is related to a higher prevalence of scores on the Armed Forces Qualifications Test (AFQT), other underclass behaviors. But in the results so far, there more years of schooling, and are less likely to be high does not seem to be a consistent pattern in this relationship school dropouts. These findings are strongest for whites across racial groups and for both sexes. Furthermore, the and blacks. For Hispanics, an increase in mothers' educa- empirical studies reported here do not tell us whether the tion is associated only with the higher AFQT scores. relationships are associational or causal. Although the theo- retical work of Wolf and his associates (paper 10) provides Corcoran and her coauthors in their study of men's earnings three or four possibilities (e.g., explanations of how neigh- (paper 1) state, "One of our strongest results is the large borhood concentration of some characteristics might influ- negative association between son's outcomes and welfare ence an individual's attitudes, preferences, or behaviors), receipt in his family of origin."*' This is not quite analogous the empirical work tells us very little about the mechanism to persistence for women, but it is close. that brings about the changes. Mark Testa and Marilyn Krogh also do not directly address welfare persistence in their analysis of nonmarital parent- Persistence of underclass characteristics hood, male joblessness, and AFDC participation in inner- Peter Gottschalk (paper 2) addresses intergenerational city Chicago (paper 6). However, in computing the relative transmission of welfare dependency. He confirms earlier risks of first AFDC receipt, they look at whether the work that showed that the grown-up children of welfare subject's family received welfare when she was a child. In mothers are more likely to be on welfare than the children general, they find that, for various racial-ethnic groups, the of nonwelfare mothers. He worries, however, that earlier likelihood of a daughter participating in AFDC is higher, works have not distinguished two possible causes of this but not significantly so, if the family was on welfare when intergenerational transmission: the mother and daughter she was a child.28 share some other characteristics (for example, living in the same neighborhood) which have an influence on their joint Shelly Lundberg and Robert Plotnick deal with the oppor- receipt, or the mother's attitudes and behavior affect the tunity costs of teenage childbearing (paper 4). They also daughter's behavior. estimate the likelihood of several marital and fertility op- tions for a female aged 19 as a function of a number of Obviously, there are policy implications to be drawn from variables, including the marital status of her mother when the conclusion reached. If the fact that a woman receives the daughter was 14. In addition, they predict the probabil- welfare increases the likelihood that her daughter will be- ity of a woman being married at age 28, given her marital- come dependent on welfare, then greater attention should fertility status at age 19, as a function of a number of be paid to changing the mother's status than would be the variables, again including the marital status of her mother case if some other variable, such as the neighborhood, was when the daughter was 14.29The probability of the daughter causing this intergenerational relationship. If some other being married at age 28, regardless of marital-fertility sta- factor causes the correlation, however, then working with tus at age 19, is generally related to mother's marital status the mother may not affect the behavior of the child. Even if when the daughter was 14. Black and white daughters of the association is confirmed between mother's welfare single mothers are less likely to be married than daughters dependence and subsequent dependence of her child, of married couples. Gottschalk points out that we must learn by what means dependence is established. The mechanisms discussed by A new paper by Anne C. Case and Lawrence F. Katz, "The Wolf and his colleagues come to mind: information shar- Company You Keep,"30 is germane to this discussion. It ing; adaptive imitation, which could include lower self- presents hypotheses which link neighborhood concentra- esteem and ambition; or peer pressure.26 tion of various behaviors and intergenerational persistence in a way that seems consistent with the approach used in the Hill and O'Neill (paper 3) also find persistence in welfare conference papers. Case and Katz discuss how both family receipt across generations. For example, in descriptive and neighborhood (measured primarily in terms of peers terms, white young women in the NLSY from welfare and local adult role models) affect behavior and outcomes families have a 24 percent chance of ever being on welfare; of disadvantaged youths in Boston. Using the Boston from a nonwelfare family the probability is 2 percent. Com- Youth Survey of 1989, carried out by the National Bureau parable figures for blacks are 42 percent and 15 percent; for of Economic Research (NBER), the researchers look at Hispanics, 34 percent and 8 percent. Furthermore, receipt three neighborhoods-South Boston, Dorchester, and of welfare is found to be significantly associated with bad Roxbury. They conclude that "different family background outcomes-dropping out of school, going to jail, increased variables have quite distinct relations with the different proportion of years with little work, and childbearing out of measures of socioeconomic outcomes for the youths in the wedlock-for both black and white young women. sample. In particular, family background variables appear to be most strongly related to similar variables for youths For several variables unrelated to welfare, Hill and O'Neill and usually not significantly related to other outcome vari- also report intergenerational persistence. They find that ables when directly related family background variables are included in the specification. In other words, youths who linked to "like behavior" in those associated with that per- had family members in jail when they were being raised are son. The three mechanisms discussed by Wolf, Clark, and much more likely to be involved in criminal activity; those Freedman-information, adaptive imitation, and peer pres- with family members with drug problems are more likely to sure-may all operate to influence people's (ir)rational use drugs; those with teenage mothers and parents who choices. As yet we do not know which mechanism, if any, were not married are substantially more likely to have is more influential. These factors may have more influence children out of wedlock; and those with more-educated on youth than on adults. Underclass neighborhoods may be parents get more ~chooling."~~The impact of family back- formed by selective out-migration of the most able. Those ground variables appears more complex than simply stating left may be least able, and they may undervalue their (lim- that "good" families have "good kids and "bad" families ited) opportunities because they cannot assess what can be have "bad" kids. They look at behaviors such as crime, obtained through traditional behavior. illegal drug use, out-of-wedlock childbearing, idleness, school completion, and church attendance. After control- Because most of the conference papers were written from a ling for family background, they find that neighborhood behavioralist perspective, sufficient weight has not been peers with a high prevalence of a particular behavior appear given to the structuralists' view of the importance of struc- to influence youth behavior of the same variety, but neigh- tural forces (e.g., the availability of education and jobs) or borhood adults have less influence. They conclude that their concerns about the changing pattern of the industrial their results are consistent with Wilson's emphasis on black sector. working-class and middle-class role models, but find that the influences on youth are channeled through the actions Patterns of simultaneous underclass behaviors of their peers. Although numerous behaviors are associated with the Hill and O'Neill (paper 3) seem to verify persistence in underclass, single-parent families, nonmarital childbear- analogous behavior from one generation to the next. They ing, and welfare uSe are the only set of behaviors addressed use a number of control variables for both young men and by a significant set of the papers. Lundberg and Plotnick; women concerning their parents (e.g., family structure and Testa and Krogh; Sullivan; Hill and O'Neill; Gottschalk; schooling for males, and family structure, welfare receipt, and Darity and his coauthors all discuss various aspects of and schooling for females). I looked at their significance the fertility-marriage-welfare behavior complex. Only this tests and found that for blacks, Hispanics, and whites, both constellation of behaviors will be covered in this synthesis. male and female, relationships were much more likely to be significant for "like-variablesw-about two-thirds of the Nonmarital births are highly associated with AFDC receipt, time-than for closely linked "similar" variables, such as long-term welfare dependency, and persistent poverty. Al- when the mother headed a single-parent family and the though at any point in time only 10 to 20 percent of AFDC daughter used welfare. These were significant three-fifths recipients are teen parents, about one-half of the recipients of the time. Relationships between "unlike variables" (e.g., had a child when they were in their teens. (One does not mother's schooling and daughter's use of welfare) were have to look at single parenting as a moral issue or dysfunc- significant only about 15 percent of the time. On the other tional behavior, although many people do. Mamage is hand, when I looked at all the significant results (about one- apparently one of the most powerful poverty fighters, quarter of the cases), they were about equally divided although I know of no research that shows how government among the three comparisons of behavior: like, similar, and can literally influence individuals to marry one another in unlike. order to combine resources to reduce poverty and welfare dependence.) David Sjoquist, in his review of the underclass literature in the paper by Martinez-Vazquez and Saposnik (paper 9). The number of births to teenagers declined from 570,000 in uses a different measure of persistence: not persistence 1977 to 470,000 in 1987, primarily because of the decline from one generation to the next but length of time or in the number of teenagers. The birthrates declined only percentage of available time that the subject exhibits the slightly, from 52.8 to 5 1.1 per thousand young women aged behavior in question. He concludes that "although the 15-19. The abortion rate increased during the mid-1970s underclass may be persistently poor, the persistently poor but has remained at about 40 per thousand young women are not necessarily members of the under~lass."~~ since then. About 60 percent of pregnancies end in births. Births outside marriage have risen from 30 percent of all This research on both neighborhood concentration and per- births in 1970 to 49 percent (1980) to 64 percent (1987). sistence of underclass characteristics seems to be consistent Although it appeared for a while that sexual activity among with an ecological human development perspective. Neigh- teenagers had leveled off in the 1980s, recent data indicate borhood influences operate in ever widening circles from that it has continued to increase.33 the local to the global. People are most influenced by fam- ily, close friends, and peers, other neighbors and adults, and When we look at welfare receipt in the underclass context, various community institutions and societal forces, in that we see (in Table 1) that nearly one-half of the black NLSY order. Furthermore, the behavior of one person seems to be women, 38 percent of the Hispanic women, and 5 percent of the white women live in areas of high concentration of ily income, controlling for other factors, has an influence welfare families. Twenty-one percent of the black women, on whether there is an out-of-wedlock birth. However, for 10 percent of the Hispanic women, and 3 percent of the blacks, coming from a single-parent family or living in white women spend a high proportion of their lives be- areas with high welfare benefits has a significant effect tween ages 19 and 27 on welfare. However, only 3 percent even after controlling for other factors. For whites, coming of the black women, 1.5 percent of the Hispanic women, from a single-parent family or a welfare family or having a and 0.6 percent of the white women are high school drop- mother with lower education is associated with higher lev- outs, unwed teen mothers, and long-term welfare recipi- els of out-of-wedlock births. ents. Testa and Krogh find that age-specific unemployment rates All the researchers find significant behavioral differences for black males living in Chicago seem to be unrelated to among women of different race-ethnicity groups, even after nonmarital parenthood but associated with the decline in controlling for class. Blacks have higher rates of preg- black marriage rates-the marriage rates for black unem- nancy, nonmarital births, and single parenting. Hispanics ployed men declined much more over time than marriage are somewhere between blacks and non-Hispanic whites. rates for employed men.35 (They point out that this finding Testa and Krogh find in Chicago that Mexican Americans supports the Wilson hypothesis-that it is lack of jobs tend to have rates more like whites, and Puerto Ricans, which has caused the decline in black marriage rates-and more like blacks. (None of the other studies created sub- contradicts the Charles Murray hypothesis-that generous groups among Hispanics, and their results in their Hispanic welfare benefits serve as a disincentive for young, em- analyses often were not significant. A distinction between ployed males to marry the mothers of their children. On the Hispanic groups might help these analyses.)34 other hand, the Hill-O'Neill and Lundberg-Plotnick find- ings related to welfare payment levels support Murray, at Looking at this set of outcomes from a rational choice least as far as white women are concerned.) Testa and perspective, one can examine the various branches of the Krogh report, however, that the decline in black male em- decision tree: sexual activity, contraception, pregnancy, ployment is still insufficient to explain the even steeper abortion, birth, marriage. Mercer Sullivan (paper 5) finds decline in black marriage rates. Nor can they explain the that poor whites, blacks, and Hispanics in New York City significant racial differences. Like Sullivan, they find have quite different patterns. The poor are less likely than higher rates of legitimization for Mexican Americans and the nonpoor to carry a child to term and less likely to non-Hispanic whites than for blacks. But they also find that legitimate births through marriage. Poor Hispanics are less black women who are in school or employed are less likely likely than non-Hispanics to use abortion and more likely than those who are not so engaged to bear a child. than blacks to get married. Lundberg and Plotnick find in their national sample of youth that whites appear to act in a Darity and his coauthors (paper 7) posit a somewhat differ- manner most consistent with the rational opportunity-cost ent structural conceptualization of the rise in black female- model often posited by economists. They tend to be more headed families from that of Wilson, although they and sensitive to incentives and costs than blacks or Hispanics. Wilson both attribute the declining black two-parent family Those whites who live in states with higher welfare benefits to the decline in marriageable black males. They focus on tend to carry their babies to term and not legitimate them. the decline in marriageable males due to imprisonment, They are less likely to use family planning and abortion in murder, suicide, drugs, and unemployment. They measure states with laws restricting birth control and abortion. How- the effects on family structure of welfare rates and ratios of ever, although being a single parent as a teen does have marriageable men to women. They find, as have others, a negative consequences for white women in terms of lower positive association between welfare use and welfare levels wage and marriage rates when they are in their late twen- (welfare use grows with increases in benefits), an associa- ties, these prospective opportunity costs do not seem to tion that is higher for blacks than for whites. They point influence their childbearing behavior. Lundberg and out, however, that since real welfare benefits declined over Plotnick do not use family income in their equations, so the period of their analyses-1 976 to 1985-increased ben- their results cannot be directly compared with Sullivan's efits cannot explain the increase in single parenting. They New York City data. They do find, however, that both also find that family structure is related to the ratio of whites and blacks whose mothers had higher levels of marriageable males to unmarried females. Whereas Wilson education are significantly more likely to terminate preg- sees the decline in marriageable black males as resulting nancies with abortion. from spatial and skills mismatches between jobs and black males, Darity and his coworkers attribute the decline to Hill and O'Neill do not look at the decisions leading up to discrimination against black males. having out-of-wedlock children. Using the same NLSY data base but a different type of analysis, they confirm the They conclude that while class matters, race matters also. Lundberg-Plotnick findings, however, that whites, but not They assert that when conventional empirical modeling blacks or Hispanics, are more likely to have out-of-wedlock finds a race effect, scholars tend to attribute it to different births if they live in states with higher AFDC and Food choices available in response to racially differentiated op- Stamp benefits. Unlike Sullivan, they do not find that fam- portunities, incentives, and disincentives. They believe that the race variable reflects more than individual choices, how to attain these norms if they are accepted. The however; it represents historical structural forces-group process of social research thus involves model building, differences-related to a history of discriminatory laws and hypothesis formulation and testing, and general applica- practices. In conclusion, Darity and his colleagues reluc- tion of the scientific method. These traditional methods tantly assert: "Central to an understanding of the rise in are well known and widely discussed. Policy analysis female-headed families and the escalation of violent crime and evaluation research are akin to them, yet quite dif- among blacks in America is the diminishing usefulness of ferent. black labor and the resulting marginalization of blacks at The key characteristic of policy analysis and evaluation the bottom of society."36 research is a focus on the activities of the public sector and their impact on the larger society. Policy analysis Generalizations about simultaneous underclass behaviors and evaluation research respond to a conscious effort by are very difficult to discern from these studies. This is due, government to change behavior or performance by to a great extent, to the large number of independent and means of public policy; they involve the examination dependent variables examined, which in turn reflect both and measurement of the impact of that policy on human the complexity of the problem and the researchers' disci- well-being, behavior, and performance. While the goal plinary training. Difficulty in synthesizing results increases of standard social science research is to advance the geometrically when one starts trying to connect multiple discipline, the primary goal of policy analysis and evalu- dysfunctional behaviors, like not completing high school, ation research is to provide information to policymakers drug use, crime, low employment rates, etc. We need a on the impact of public measures designed to change systematic way to tally the results. behavior or outcomes.3s

Nevertheless, it seems apparent from these studies that race This is a very important distinction. Positive social science does make a difference, which cannot be accounted for by is about understanding behavior in the environment as it is. controlling for a number of other economic, social, and Policy analysis is about understanding how we might community variables. It also appears that there may be change the environment to achieve policy goals. Positive intergenerational persistence of like behaviors: daughters social science is a necessary but insufficient condition for of single or welfare parents are more likely than daughters policy analysis based on research. For effective policy pre- from other types of families to be single and on welfare. scription we need to know not only that behavior (B) is Unfortunately, these two conclusions, like much of this associated with group (G) or that social force (F) leads to research, tell us little if anything about why this is the case level (L), but whether treatment (T) to B or F can move G or and what government policy might do to alter underclass L in the "right" direction. behaviors or reduce the number or size of neighborhoods with large numbers of underclass people. The research on the underclass represented by these papers can be characterized as positive social science. Some of the authors and discussants alluded to public policies or dis- Avenues for future policy research cussed the importance of knowing which of several causal routes might explain results, since different public policy The role of ASPE and other government agencies prescriptions would follow, depending on the causes of What does this research tell policymakers, if anything, underclass behavior and circumstances. about what government can do to ameliorate problems associated with the underclass? Given the lack of agreement on the definition of the under- class or its causes and complex nature of the issue, it seems said in 1969 that "the role of to me that research on the underclass throughout this de- social science lies not in the formulation of social policy, cade will continue to be positive social science rather than but in the measurement of its results."37 Robert Haveman policy analysis. I do not foresee a large investment of recently made the distinction between two types of social government resources in demonstration evaluations or research: (I) objective social science research and (2) more other forms of applied social science related to the under- normative policy analysis and evaluation research. He class qua underclass, although research will be conducted stated that on particular aspects of underclass behavior such as teen parenting or welfare dependency. Any serious discussion Traditionally, research in the social science disciplines of the policy implications from underclass research that I has been "positive" in its orientation. That is, basic have reviewed is, in my opinion, very premature, although social science research has sought to understand the understandable given the severity of the problem. While world of human and social behavior. The question posed social scientists can be accused of studying a problem to is, What is the nature of social behavior, and can those death, premature initiatives also have their costs and unin- factors which influence and determine it be identified tended consequences. (I applaud the efforts of the Urban and their impact measured? This approach emphasizes Institute and Ford Foundation for their Urban Opportunities the positive issue of understanding what is; little atten- Program, which is trying to develop a demonstration tion is given to establishing norms for what should be, or agenda.)39 It is likely, however, that government agencies will con- "Unfortunately, models which emphasize major differ- tinue to investigate underclass issues as articulated by ences by class and the large societal forces that create and Jencks in the conference discussion and in his writingsM- shape such classes could not be included [in his analysis] educational underclass, jobless underclass, impoverished because they cannot be as easily subjected to traditional underclass, etc. Separate underclass behaviors are the prov- tests, which focus on individual behavior" (p. 6, emphasis ince of a number of federal departments: dropping out of added). That is, as an economist, Ellwood had trouble school, the Department of Education; crime and delin- evaluating sociological paradigms, which he openly admit- quency, the Department of Justice; out-of-wedlock child- ted at the conclusion of his article. bearing and welfare dependency, the Department of Health and Human Services; and concentration of the poor, the In response to Ellwood, Sullivan, an ethnographer, pro- Department of Housing and Urban Development. There- vides another perspective of rational choice and culture. fore, more interdepartmental collaboration is also neces- His paper describes how rational choice and cultural factors sary. Foundations and other institutions can play an impor- interact in determining patterns of childbearing, marriage, tant role in promoting collaboration and pulling together and AFDC use. He argues that culture has been mischarac- various funding strands. terized as a set of static traits passed on unreflectively from one generation to another.43 He states, "contrary to Ell- Government research institutes, like the National Institutes wood's insistence on separating out 'pure models' of of Health, and statistical agencies, like the Bureau of the choice and culture, this paper has shown that culture itself Census, tend to have research portfolios that are primarily contains a substantial element of rational choice. Choices invested in positive social science, whereas a policy shop, can be individual cost-benefit calculations but they can also such as ASPE, tends to have a portfolio which is more be collectively patterned solutions to common problems. oriented to normative policy analysis and evaluations. In This paper vindicates many aspects of all three kinds of judging where to invest scarce resources for research and theory, with the notable exception of variants of culture evaluation, ASPE emphasizes policy relevance and timeli- theory which treat culture as a set of unchanging values not ness to inform p~licymaking.~' responsive to the changing exigencies of adaptation to cir- cumstance~."~~ It is my opinion, therefore, that ASPE should invest only modest amounts in research on the underclass. The prepon- The vast majority of data used in social science in general derance of funding for such research should be provided by and underclass research in particular are survey data. Un- foundations and academic institutions until such time as fortunately, each discipline approaches a survey differ- program policy options come clearly into focus. On the ently. They have different data demands and survey strate- other hand, a case can certainly be made for ASPE to gies. Given both serious research budget constraints and support basic research tools, such as the NLSY and Panel individual survey time constraints (for example, the one- Study of Income Dynamics (PSID) surveys, upon which hour time limit guideline for an interview used by the researchers depend in studying the underclass and other Office of Management and Budget), almost insurmount- poverty problems. able interdisciplinary problems arise.45It appears to me that as long as surveys are an important source of data, interdis- Regardless of whether ASPE should invest current budgets ciplinary collaboration is imperative to bring about the in underclass research, the problem is serious and does effective use of limited resources. need further attention. What seem to be promising ave- nues? The next section discusses two strategies-multidis- Surveys, however, are not enough. Those now in use may ciplinary/multiple-data approaches and longitudinal analy- be seriously flawed for research on the underclass. No sis. The final section lists several open questions that seem matter how you define the underclass, it includes a high important. prevalence of nontraditional behaviors in addition to pov- erty. Darity and his colleagues point out that surveys such Strategies as the PSID or NLSY, which follow respondents over a long period of time, may suffer serious in under- Multidisciplinary approaches. How can we support data representing the underclass and underreporting underclass gathering which will encourage multidisciplinary para- behavior because underclass members are less likely than digms and analyses? Is it possible? Will it lead to more others to be selected into the sample frame for initial inter- cost-effective research? viewing and are more likely to drop out of a survey over Economists study how people make choices; sociologists time. They make the further point that all groups are likely study how people have no choices to make. (anonymous) to underreport underclass behavior. It does appear, there- fore, that traditional survey methods will have serious limi- In an earlier issue of Focus, David Ellwood analyzed three tations in obtaining valid, reliable information about models-rational choice (choices and behavior), expect- underclass phenomena, no matter whether the paradigm ancy (confidence and control), and cultural (values and stresses attitudes, expectations, behaviors, or social forces. culture).42He stated these roughly correspond to the disci- That does not mean surveys should be discarded. They plines of economics, social psychology, and anthropology. must be supplemented with other types of data. I believe Wilson had a bold research strategy in his Urban and out of those neighborhoods. They inferred from the Family Life (LTFL) study by attempting to blend a survey, cross-sectional data that underclass people were stayers ethnographic case studies, research of institutions, and sec- rather than movers. We have yet to see dynamic analyses of ondary analysis of administrative records and historical the underclass of the sort that has been done on people in data. His plan to include eight or ten ethnographers in poverty or on welfare rolls. different neighborhoods to gather a common set of infor- mation was an important supplement. No doubt there is The works of Greg Duncan, Saul Hoffman, Mary Jo Bane, some validity to concerns about ethnography; case studies, and David Ellwood using the PSID on the dynamics of like journalistic anecdotes, are difficult to replicate and poverty and, later, on welfare dependency are credited with validate. That is why Wilson's strategy of having a number changing the way people look at these problems.4s They of ethnographers gather data from informants on a number showed, for example, that poverty was much more dynamic of questions relating to numerous aspects of the underclass than had been assumed. Looking at completed spells over a and guided by a single (or small group acting as a) control- number of years, they found that more people experienced ler has a great deal of appeal to me. It is my understanding some poverty than had been previously shown, and, for that the controller was to assist the ethnographers in com- most people in poverty or on welfare, the average length of paring notes and hypotheses and to force them, as a devil's stay was short. For those who made up the long-stayers, advocate, to challenge preconceived notions and assump- however, the stay was longer than they had expected; at any tions about the poor. On the other hand, it is my impression single point in time a significant proportion of welfare that most ethnographers would be unhappy with such a recipients are in the midst of a long spell of welfare depen- structured approach; it would conflict with standard ethno- dence. This longitudinal perspective represents a major graphic techniques. shift in our view of a social problem-a change in the life cycle of an issue-from static description to a more dy- This approach seems to me, however, to overcome many namic frame of reference. limitations that other social scientists see in ethnography. Yet, having read a number of the ethnographic case reports We seem to be close to the same developmental time phase from the Wilson study, I find it very difficult to conceive in research on the underclass. Longitudinal analysis of how these extremely rich and provocative anecdotes could underclass concentration, persistence, and behaviors will be organized and integrated with the project survey or other no doubt shed new light on this issue. It is possible that just data in a systematic manner. I await the reports from the as did Sharon Long, who looked at hazard analysis of researchers to see if they accomplished the integration of multiple program receipt, researchers will be able to ex- qualitative and quantitative data. plore patterns of multiproblem underclass behaviors.49

Given that labor force participation is central to the The work anticipated from Corcoran, which merges infor- underclass issue, looking at historical trends in jobs and mation about the census tract area where PSID respondents employment and interviewing employers also seem critical. have lived, and from Hill and O'Neill, which merges zip Economists talk about rational choice in supply and de- code information with NLSY respondents, will allow us to mand. Sociologists are concerned with institutional re- look at people over time and link them to their neighbor- sponses that limit individual options. It seems to me that the hoods. These studies and others should provide a much available analytical techniques favor looking at individual more dynamic definition and analysis of the underclass choice rather than institutional behavior, whether it is de- (concentration, persistence, and behaviors). Thus far, how- mand for workers or supply of educational opportunities. ever, these works have not investigated in- and out-migra- (All of the conference papers use data on individual behav- tion of underclass people from neighborhoods. ior rather than institutional response to individuals; this is probably also representative of a general bias in ASPE- funded research.) More can be done. Wilson's colleagues Issues needing further research interviewed employers about hiring practices and, I be- What follows is a short, eclectic list of issues. It is not lieve, obtained some very important insights.46 They also intended to be a complete, definitive list of need-to-dos. As obtained and analyzed employment data for Chicago over a I mentioned earlier, I believe that our understanding of the period of years. underclass is insufficient at this point to allow us to attempt applied policy research or program evaluation. It is important to our understanding of the underclass, in my opinion, to formulate interdisciplinary strategies to blend 1. Single parenting, marriage and unemployment. What is several data sources and to look at both individuals and the causing the decline in the rate of marriage among adults? institutions with which they interact. Why is there an even steeper decline for blacks than for whites or Hispanics? Why is there such an increase in Longitudinal analysis. To date, most of the research re- children being born to and raised by single parents? Is there ports on neighborhood concentration of the underclass have really a wage premium to a black teenager to have a child, analyzed one or two cross-sectional surveys4' but have not as found by Lundberg and Plotnick? Are employment pros- looked at longitudinal microdata on individuals flowing in pects for males an important factor? Some policy analysts feel tkat~~marriageis not an appropri- 2. How is membership in the underclass transmitted? It is ate topic for government policy research. Perhaps not, since not obvious to me that just because children live in public policy probably should take a neutral position that underclass areas they should be counted as members of the is neither pro- nor anti-marriage in individual situations. underclass. If one is willing to grant that children are not Value judgments aside, it does seem to me, however, to be born into the behavioral underclass, what mechanisms en- an important positive social science research topic, espe- courage dysfunctional behavior in adolescence and adult- cially since many public policies have unintended conse- hood? How are some low-income families in underclass quences to children through the breakdown of marriage and areas able to avoid underclass behaviors? It also is not child support. In a recent survey, 30 percent of the members obvious to me that underclass behavior occurs only in of the American Psychological Association found the de- metropolitan areas. It may be that the emphasis on the cline in the nuclear family to be the single greatest threat to underclass in cities has to do with the data available. What America's mental health-ahead of unemployment (20 about rural areas? Is the density of the population a contrib- percent), drug abuse (18 percent), and alcohol abuse (14 uting factor, or are there underclass people or areas in percent).50 nonmetropolitan regions? Is it the people in the neighbor- hood or something about the geography that contributes to Testa and Krogh's work on the relationship between unem- underclass behavior? ployment rates and marriage rates is an important issue in my opinion. Since their work is only applicable to Chicago Assuming there are many factors related to underclass be- and may suffer from selection bias problems, it must be haviors, what are their relative contributions? Do they dif- replicated by other studies. It would be of great value to fer for different behaviors, and if so, how? It seems to me know whether welfare causes increased nonmarital parent- that regardless of who influences underclass individuals- hood among the employed (as Murray has suggested) or their parents (intergenerational persistence) or their friends, unemployment prevents men from marrying and therefore peers, and neighbors-we eventually need to understand is related to single parenting and dependence on welfare (as the causal mechanisms in order to move from positive Wilson suggests). Robert Moffitt, in his recent review of social science to policy analyses and prescriptions. The the literature on the incentive effects of the U.S. welfare formulation by Wolf and his colleagues of how concen- system, also concludes that there should be more research trated behavior influences others is important. Gottschalk on family str~cture.~'A (positive social science) finding states that in order to develop policy prescriptions we still that high unemployment of black males is related to high need to know why children follow in their parents' foot- out-of-wedlock childbearing and single-parent families, steps. Lundberg and Plotnick assume that younger cohorts however, does not imply that employment programs for perceive opportunity costs of out-of-wedlock childbearing males will increase marriage and reduce single-parent from the general experiences of an older national cohort-a families. That hypothesis would require testing with a cohort that did not necessarily even live in the younger (policy research) demonstration; marriage rates would be cohort's neighborhood. Possibly one of the reasons they do an important variable to measure in addition to standard not find apparent opportunity-cost effects is that their for- employment and training variables when and if male em- mulation does not take account of the lack of proximity of ployment demonstrations are undertaken. role models to the younger cohort or the manner in which their tastes are formed or modified.

A natural experiment apparently took place in Boston from 3. Criminal behavior. Definitions of the underclass often the late 1970s until recently-the tremendous economic include criminal behavior. Much of the empirical work, boom and tight labor market. Maybe the NBER survey used however, has not included it. Is this situation due to limita- in the work of Richard Freeman, Case, and Katz can add to tions in data or parochial disciplinary training and failure to our understanding of the relationship between the labor design broader paradigms? Can criminal behavior, includ- market and family structure.52Early results indicate that the ing substance abuse, be fruitfully linked to (behavioral and economic boom did reach down into the ranks of low- structural) paradigms and analyses of other behaviors and income black males. Has it affected family formation? neighborhood concentration?

Paul A. Jargowsky and Mary Jo Bane report a significant The NLSY, Wilson's survey, and ethnographic data present reduction in ghetto areas in smaller metropolitan areas in such an opportunity. Richard Freeman's Boston survey the So~th.~'These cities might serve as natural experiments also includes data on criminal behavior. His recent work to look at employment and marriage rates where circum- reports the terrible consequences that incarceration has on stances have gotten better, as well as worse. We might be prospects for future empl~yrnent.~~These results are con- able to look at these successful labor markets, in addition to sistent with the Darity et al. concept of the marginalization Boston, instead of focusing entirely on the problems of of black men. rustbelt cities, as has been much more the case to date. Can we bring longitudinal, dynamic analyses to bear to help us 4. The role of welfare. The data from these papers seem to understand whether employment policies have positive im- indicate that a family's welfare use is associated with both plications for families? intergenerational persistence and neighborhood concentra- tion of welfare families. What mechanism(s) is (are) driv- operationalized. That is, he did not lay out details of his theories in ways ing these phenomena; or are they just spurious correla- that one could test with empirical data. In his 1990 presidential speech at the American Sociological Association, one of the reasons he gave for tions? Can public administration reverse these conse- encouraging the work of Van Haitsma was that her theoretical work is quences? If so, how? If higher welfare payment levels embedded in a framework of testable hypotheses. increase the prevalence of welfare and lower levels raise Van Haitsma's colleagues Mark Testa and Marilyn Krogh made a contri- the poverty rates, what policy demonstrations should be bution in this regard in their ASPE work (paper 6, see box, p. 7). They undertaken? took two major theories about the underclass-Wilson's and Charles Murray's-and discussed what they thought would be empirical tests of This list of issues outlines some of the many unknowns we the two hypotheses as they relate to male unemployment and marriage face. We still do not have agreement on what we mean by rates. They used the discussion of the contending hypotheses to guide the presentation of their results. However, since Wilson and Murray often the underclass, much less what factors are associated with did not articulate their theories in empirical terms, people can disagree its growth. We do not know whether government can re- about whether Testa and Krogh have specified their hypotheses cor- duce the causes of the growth of the underclass and if such rectly. This is yet another example of the frustrations of trying to pin a reduction would affect the number of people so classified. down the underclass concept. Indeed, the debates about the underclass sound very famil- I1see, for example, Paul Jargowsky and Mary Jo Bane, "Neighborhood iar to those who were around in the 1960s, when the culture Poverty: Basic Questions," Center for Health and Human Resources of poverty held center stage. About that debate Henry J. Policy, John F. Kennedy School of Government, Harvard University, March 1990. Aaron said, 12For examples of areal/spatial definitions see Mincy, Sawhill, and In retrospect, the debate between the cultural and envi- Wolf, "The Underclass"; Ricketts and Sawhill, "Defining and Measuring ronmental views of poverty seems to have vanished the Underclass"; Mark Alan Hughes, "Concentrated Deviance and the without leaving significant residue. The rea- 'Underclass' Hypothesis," Journal of Policy Analysis and Management, son may be the failure of either side in the debate to 8 (1989). 274-282; M. Anne Hill and June O'Neill (paper 3, see box, p. 6); and Mary Corcoran. Roger Gordon, Deborah Laren, and Gary formulate the issues precisely, the lack of evidence . . . Solon (paper I, see box, p. 6). and the unwillingness of participants to suggest the kinds of tests or information that would resolve the 13~thnographerssuch as Mercer L. Sullivan (paper 5) and Eli Anderson, Sullivan's discussant. are more comfortable with social concepts of debate.55 community than spatial designations of neighborhoods using census tracts or zip codes. Sullivan's paper does use a methodological approach What residue this underclass debate will leave remains to that links ethnographic case studies with health statistics. be seen. 14see David L. Sjoquist, "Concepts, Measurement, and Analysis of the . Underclass: A Review of the Literature," pp. 1-9 (included in paper 9). 15For example. see Mincy, Sawhill, and Wolf, "The Underclass." 16~heNLSY does not represent all age groups; it only represents people l~heedited proceedings of that conference have been published as The who were 14 to 21 years old in 1979. However, they were 22 to 29 years Urban Underclass, ed. Christopher Jencks and Paul E. Peterson (Wash- old when the analysis was done. This age cohort is a very significant ington, D.C.: The Brookings Institution, 1991). portion of the underclass and therefore informative. 2~heauthor would like to thank Katherine McFate, Dan Meyer, Sharon 17~arityand his coauthors (paper 7) struggle with this problem in the McGroder, and Steve Sandell for their helpful comments and sugges- first chapter of their paper. tions and the Policy Information Center library staff for their help in preparing the study abstracts. 18"~tatisticalDiscrimination and Inner-City Workers: An Investigation of Employers' Hiring Decisions," University of Chicago, August 1990. 3~illiamA. Darity, Jr., Samuel L. Myers, Jr., William J. Sobol, and Emmett Carson (paper 7, see box, p. 8), pp. 1-7. 19For example, see "The Dangers of the Underclass: Its Harmfulness as a Planning Concept," Russell Sage Foundation Working Paper no. 4, New York, January 1990. 5~heTruly Disadvantaged: The Inner City, the Underclass, and Public 20~eeWilson, "Social Theory and Public Agenda Research: The Chal- Policy (Chicago: University of Chicago Press, 1987). p. 8. lenge of Studying Inner-City Social Dislocations," presidential address, 6''~Conceptual Definition of the Underclass," Focus, 12: 1 (Spring and annual meeting of the American Sociological Association, Washington, Summer 1989), p. 28. D.C.. August 1990. 7~ee,for example, Ricketts and Sawhill, "Defining and Measuring the 2'~onferencetranscript, p. 257. Underclass," Journal of Policy Analysis and Management, 7 (Winter 22~onferencetranscript. p. 194 1988), 316-325; Ronald B. Mincy, Sawhill, and Douglas A. Wolf. "The Underclass: Definition and Measurement," Science, April 27, 1990. pp. 23~eeWolf. Clark, and Freedman (paper 10). pp. 4344. 45M53. 24~eeManski. "Dynamic Choice in a Social Setting," SSRI Workshop 8~incy,Sawhill, and Wolf, "The Underclass," p. 451 Series. University of Wisconsin-Madison, March 1990; and Streufert, "Middle-Class Flight and the Underestimation of Education's Future 9~ee.for example, Douglas S. Massey. Mitchell L. Eggers, and Nancy A. Benefit by the Underclass," Institute for Research on Poverty proposal to Denton, "Disentangling the Cause of Concentrated Poverty," Population ASPE for the 1989-1991 biennium. Research Center, University of Chicago, October 1989. 25~orexample, see Massey, Eggers, and Denton, "Disentangling the 10~ilson'searlier theoretical work was hampered, in my opinion, by the Cause of Concentrated Poverty"; and John D. Kasarda, "Urban Industrial fact that in some parts of his theory the hypotheses were not stated in Transition and the Underclass," The At~t~alsof the American Academy of refutable terms or it was not clear how the variables could be Political and Social Science, January 1989, pp. 2647. 26~ottschalk'spaper was primarily methodological in content. Greg 51"~ncentiveEffects of the U.S. Welfare System: A Review," Institute Duncan's critique included a number of small and large methodological for Research on Poverty Special Report no. 48, Madison, Wis., 1991. concerns. See conference transcript, pp. 12 ff. 52~ichardFreeman, "Crime and the Economic Status of Disadvantaged 27~.12. Young Men," presenled at the Conference on Urban Labor Markets and Labor Mobility, Urban Opportunities Program, the Urban Institute, 28~estaand Krogh, Tables 8.24, 25, and 26. Washington, D.C., March 1991; Case and Katz, "The Company You 29~undbergand Plotnick, Table 6.1, p.40; and Table 6.4 A-C, pp. 49- Keep." 51. 53"~eighborhoodPoverty: Basic Questions." 30"~heCompany You Keep: The Effects of Family and Neighborhoods 54~reeman,"Crime and the Economic Status of Disadvantaged Young on Disadvantaged Youths," NBER Working Paper no. 3705, May 1991. Men." 31~bid.,pp. 2-3. 55~oliticsand the : The Great Society in Perspective (Wash- ington, D.C.: The Brookings Institution, 1978), p. 22. Quoted by Darity et al. (paper 7), p. 21. 33~hesefigures are taken from comments by Kristin Moore in her discussion of the Lundberg-Plotnick paper. 34~orexample, see Marta Tienda and Haya Stier, "Joblessness and Shiftlessness: Labor Force Activity in Chicago's Inner City Neighbor- hoods," in The Urban Underclass, ed. Jencks and Peterson. 35~onaldFerguson, one of the discussants, correctly points out that this may be related to selective out-migration from the sample neighbor- hoods by older males. Testa acknowledges this possibility in his paper. 36~arityet al., Executive Summary, p. i. 37~akenfrom Richard P. Nathan, Social Science in Government: Uses and Misuses (New York: Basic Books, 1988). p. 14. 38~obertH. Haveman, Poverty Policy and Poverty Research: The Great Society and the Social Sciences (Madison, Wis.: University of Wisconsin Press, 1987), pp. 156157. 39~heUrban Opportunities Program is designed to look at the changing structure of American cities and the implication for economic and social opportunity. The project is being carried out by the Urban Institute under Ford Foundation sponsorship. For additional information contact George E. Peterson or G. Thomas Kingsley at the Urban Institute, (202) 833- 7200. 40~ee,for example, "What Is the Underclass-and Is It Growing?" Focus 12:l (Spring and Summer 1989). pp. 14-26. Order forms for Focus, Insights, and 41~hetypology described here is very similar to a three-part typology other Institute publications are discussed by Nathan, Social Science in Government, p. 14. 42"~heOrigins of Dependency: Choices, Confidence, or Culture?" Fo- at the back. cus, 12:l (Spring and Summer 1989), pp. 6-13. Subscribe now to our Discussion Paper 43~onferencetranscript, pp. 73-74. 44~ullivan(paper 5), p. 38. Series and Reprint Series. 45~once was involved in a research project that required designing a Please let us know if you change questionnaire to gather survey data about the incidence of running away. I found that the sociologists and psychologists had completely different your address so we can continue to perspectives on how questions should be asked and in what order they should be asked. It is my judgment that a great deal of this conflict was send you Focus. the result of academic paradigms, rather than the idiosyncracies of the individuals involved. 46~eckermanand Kirschenman use these interviews in "Statistical Dis- crimination and Inner-City Workers." 47~orexample, see Ricketts and Sawhill, "Defining and Measuring the Underclass." 48~orexample, see Bane and Ellwood, "Slipping Into and Out of Pov- erty: The Dynamics of Spells," Journal ofHuman Resources, 21 (Winter 1986), 1-23; and Greg J. Duncan et al., Years of Poverty, Years of Plenty (Ann Arbor, Mich.: Institute for Social Research, University of Michi- gan, 1984). 49~ong,"Children and Welfare: Patterns of Multiple Program Participa- tion," Urban Institute, Washington, D.C., September 1990. 50"The Threat of Broken Families," USA Today, March 19, 1991